The North American Monsoon

Page created by Marvin Fleming
 
CONTINUE READING
The North American Monsoon
                                                                         David K. Adams and Andrew C. Comrie
       Department of Geography and Regional Development, The University of Arizona, Tucson, Arizona

  ABSTRACT

          The North American monsoon is an important feature of the atmospheric circulation over the continent, with a re-
      search literature that dates back almost 100 years. The authors review the wide range of past and current research deal-
      ing with the meteorological and climatological aspects of the North American monsoon, highlighting historical devel-
      opment and major research themes. The domain of the North American monsoon is large, extending over much of the
      western United States from its region of greatest influence in northwestern Mexico. Regarding the debate over moisture
      source regions and water vapor advection into southwestern North America, there is general agreement that the bulk of
      monsoon moisture is advected at low levels from the eastern tropical Pacific Ocean and the Gulf of California, while the
      Gulf of Mexico may contribute some upper-level moisture (although mixing occurs over the Sierra Madre Occidental).
      Surges of low-level moisture from the Gulf of California are a significant part of intraseasonal monsoon variability, and
      they are associated with the configuration of upper-level midlatitude troughs and tropical easterly waves at the synoptic
      scale, as well as the presence of low-level jets, a thermal low, and associated dynamics (including the important effects
      of local topography) at the mesoscale. Seasonally, the gulf surges and the latitudinal position of the midtropospheric
      subtropical ridge over southwestern North America appear to be responsible for much spatial and temporal variability in
      precipitation. Interannual variability of the North American monsoon system is high, but it is not strongly linked to El
      Niño or other common sources of interannual circulation variability. Recent mesoscale field measurements gathered
      during the South-West Area Monsoon Project have highlighted the complex nature of the monsoon-related severe storm
      environment and associated difficulties in modeling and forecasting.

1. Introduction                                                     extensive research literature has developed that deals
                                                                    with a range of meteorological and climatological as-
a. Aims and rationale                                               pects of the phenomenon. Surprisingly, no comprehen-
   The North American monsoon, variously known as                   sive review of the NA monsoon literature has been
the Southwest United States monsoon, the Mexican                    presented. Several papers have briefly summarized the
monsoon, or the Arizona monsoon, is experienced as                  basic climatology of the NA monsoon (e.g., Maddox
a pronounced increase in rainfall from an extremely                 et al. 1995), but to date, no single contribution has
dry June to a rainy July over large areas of the south-             sought to integrate the wide range of research relating
western United States and northwestern Mexico. These                to this notable feature of the atmospheric circulation
summer rains typically last until mid-September when                over North America.
a drier regime is reestablished over the region. The                   In this paper, we provide a synthesis of major cli-
North American monsoon (NA monsoon) has been                        matological and meteorological research themes re-
studied since the early part of this century, and a fairly          garding the NA monsoon: we outline the past debate
                                                                    and emerging consensus on mechanisms and moisture
                                                                    sources; we assess the state of knowledge pertaining
Corresponding author address: Dr. Andrew C. Comrie, Depart-         to the extent and variability of the NA monsoon, both
ment of Geography and Regional Development, The University
                                                                    spatially and temporally; and we include an overview
of Arizona, Tucson, AZ 85721.
E-mail: comrie@climate.geog.arizona.edu                             of research on the severe storm environment and mod-
In final form 19 May 1997.                                          eling efforts associated with the NA monsoon. We also
©1997 American Meteorological Society                               address two additional issues. First, considering the

Bulletin of the American Meteorological Society                                                                               2197
importance of the NA monsoon within the atmo-
spheric circulation over North America, and despite
several recent research papers, misunderstandings of
NA monsoon dynamics and influence are still surpris-
ingly widespread within the broader meteorological
and climatological community. Second, this is an area
of research that has become quite active, particularly
during the last few years with the South-West Area
Monsoon Project (SWAMP), and it offers many theo-
retical and applied avenues for further investigation.
We intend this paper to be a timely contribution that
will critically summarize the state of knowledge and
stimulate further inquiry on the NA monsoon.
   Pulling together the diverse array of NA monsoon
studies into a cohesive body of literature is a challeng-
ing task. We have chosen to divide the literature into
several categories, which by nature are somewhat ar-
bitrary and partially overlapping. Many of the articles
reviewed cover a number of these categories, and thus          FIG. 1. Important physiographic features, states, and place
a particular study may appear in each of several ap-        names mentioned in the text: Empalme (E), Mazatlán (M),
propriate sections. We review papers found primarily        Phoenix (P), and Yuma (Y).
in the major meteorological and climatological jour-
nals, including those that focus principally on the NA
monsoon as well as those less directly connected but        (SSTs) at middle and subtropical latitudes along the
that nonetheless provide useful insights. Also included     Pacific Coast are mainly cool on average (< 25°C), but
are a small number of books and, where necessary, less      SSTs in the Gulf of California and tropical eastern
easily available scientific reports. Some Spanish-lan-      Pacific (farther south) are warm (> 28°C). To the east,
guage references have been included for completeness,       the region is bounded by the Gulf of Mexico, with
but the reader is referred to Reyes et al. (1994) for a     warm SSTs (> 26°C), and by the central plains of the
more complete Spanish-language reference list.              United States. The interior of the region is character-
                                                            ized by several large upland areas: the Colorado Pla-
b. Climatological setting                                   teau extends north and east to the Rocky Mountains
   We use the term “North American monsoon” in              from Arizona’s Mogollon Rim; the north–south
order to be geographically all-encompassing. It is im-      aligned mountain ranges of the basin and range prov-
portant to note that most of the NA monsoon litera-         ince dominate through Nevada, southwestern Arizona,
ture has a strong geographical bias toward the south-       and northwestern Sonora; the Mexican Altiplano (pla-
western United States, especially the state of Arizona,     teau) is defined by the Sierra Madre Occidental and
necessitating frequent mention of these areas in this       Oriental to the west and east, respectively. In addition,
review (and in our suggestions for future research di-      the peninsular ranges of southern California and Baja
rections). We wish to emphasize that this phenomenon        California along with the Sierra Nevada play impor-
influences larger areas of the southwestern United          tant climatological roles for the interior deserts, lim-
States and that, in fact, it is centered over northwest-    iting penetration of marine moisture from the Pacific
ern Mexico. A discussion of areal extent is presented       Ocean. There are also two lowland areas associated
later in the paper, but we introduce the broader mon-       with the NA monsoon: the lower Colorado River val-
soon region here to provide background on important         ley and neighboring low desert areas, which play a
physiographic characteristics.                              critical role in the formation of the thermal low; and
   Figure 1 highlights the major physical and topo-         the coastal lowlands of Sonora and Sinaloa, between
graphic features of southwestern North America that         the Gulf of California and the western flanks of the
relate to the NA monsoon. The region is bounded to          Sierra Madre Occidental. Most of the lower elevation
the west by the Pacific Ocean, including the Gulf of        zones of the monsoon region, particularly the Sonoran
California. Summertime sea surface temperatures             Desert, receive up to 80% or 90% of possible summer

2198                                                                                   Vol. 78, No. 10, October 1997
insolation, and summer surface temperatures can be           times referred to as “Sonora storms” (Campbell 1906),
extremely hot, often exceeding 40°C.                         were first thought to have their origin in the Mexican
   The combination of seasonally warm land surfaces          state of Sonora, from where they would travel north-
in lowlands and elevated areas together with atmo-           westward to southern California. The proximity of
spheric moisture supplied by nearby maritime sources         Sonora to Arizona made this northward movement
is conducive to the formation of a monsoonlike sys-          plausible for convective storms in Arizona, but the
tem. With the exception of Ramage (1971), most au-           distance to California and the localized nature of the
thors regard this circulation as a “true” monsoon, based     storms once in California seemed to negate this pos-
on seasonal reversal of pressure and wind patterns,          sibility (Blake 1923). Another explanation suggested
energy and mass transfers, and characteristic regimes        that the intensification and movement of the semiper-
of rainfall and temperature (e.g., Bryson and Lowry          manent thermal low pressure over the Colorado River
1955a; Krishnamurti 1971; Tang and Reiter 1984;              valley were responsible for widespread rain in Arizona
Douglas et al. 1993). Two points are worth noting with       and California (Beals 1922). Others, too, believed that
respect to our coverage of the NA monsoon: first, sea-       a relationship existed between the intense thermal low
sonal wind direction changes over the North Ameri-           and summer thunderstorms in the region (Ward 1917;
can Arctic might technically be considered a monsoon,        Blake 1922; Willett 1940). It was suspected that the
but we exclude this feature; second, the term “North         resulting pressure gradient would draw Gulf of Cali-
American monsoon” has been applied to a more ex-             fornia moisture northward, enhancing instability and
tensive phenomenon that we do not explicitly cover,          increasing convective activity in the mountains and
which includes the wet season over Central America           deserts of the southwestern United States. Remark-
and even the seasonal displacement of the intertropi-        ably, the notion of the Gulf of California as a mois-
cal convergence zone (ITCZ) over the eastern Pacific         ture source for NA monsoon precipitation would not
Ocean, for example, in the Pan-American Climate              appear prominently in the literature again until several
Studies program (PACS 1996).                                 decades later.

2. Historical perspective

   Historically, research interest in the NA monsoon
has its roots in the peculiar distribution of summer
precipitation in this arid and semiarid region of North
America. Figure 2 illustrates monthly rainfall distri-
butions for various sites across the region, many of
which display the characteristic NA monsoon mid- to
late summer rainfall maximum following extremely
dry conditions in May and June. Early researchers
noted that July and August thunderstorms were abun-
dant in the uplands of Arizona and western New
Mexico, and relatively infrequent to the west in south-
ern California (Campbell 1906; Beals 1922; Blake
1923). However, the origin, frequency, and distribu-
tion of this warm season convective activity were not
seen as necessarily linked to some larger-scale regional
control. In fact, it was this curious distribution of warm       FIG. 2. Seasonal distribution of precipitation across southwestern
season precipitation (nearly absent in southern Cali-        North America. Note that northwestern Mexico shows the
fornia and quite frequent in elevated zones of central       strongest monsoon signal, which diminishes through Arizona,
Arizona) that spurred much of the original research          New Mexico, and Nevada. Northeastern Mexico and Texas
                                                             display early summer–late fall precipitation peaks, while the West
that would attempt to clarify the general character of
                                                             Coast has a dry summer Mediterranean distribution (vertical axis
the summer circulation of the desert Southwest (Reed         of all graphs represents 180 mm with 20 mm increments). Areas
1933, 1937, 1939). For example, the infrequent sum-          south of the broken line receive greater than 50% of their annual
mer thunderstorms of the California deserts, some-           rainfall in July, August, and September (after Douglas et al. 1993).

Bulletin of the American Meteorological Society                                                                            2199
rapid onset of rainfall in the Arizona
                                                                                        summer monsoon was defined as a
                                                                                        “singularity” (the recurrence of an at-
                                                                                        mospheric phenomenon near a par-
                                                                                        ticular calendar date), and it was found
                                                                                        to be related to a sudden shift and re-
                                                                                        adjustment in the midtroposphere
                                                                                        hemispheric circulation. This shift in
                                                                                        circulation, resulting from the west-
                                                                                        ward and northward expansion of the
                                                                                        Bermuda high sometime around the
                                                                                        end of June, causes dry southwesterly
                                                                                        midlevel winds from the Pacific
                                                                                        Ocean to be replaced by moist south-
                                                                                        easterly winds (Fig. 3). This sequence
                                                                                        of synoptic patterns places Arizona
                                                                                        under a light southeasterly flow,
                                                                                        which seemingly implied water vapor
                                                                                        advection solely from the Gulf of
    FIG. 3. Pentads of 500-mb geopotential height (dam) from 1979–88 showing the Mexico. Consequently, the Gulf of
transition from southwest flow over the NA monsoon region in early June to east or Mexico was viewed as the unique
southeast flow in July. Note rapid northward shift in height contours during last week moisture source for the NA mon-
of June and initiation of light easterly flow over northern Mexico (after Bryson and soon—a view that continued to appear
Lowry 1955b).
                                                                                        in the literature for many years (Reitan
                                                                                        1957; Jurwitz 1953; Green 1963;
                                                                                        Green and Sellers 1964; Hastings and
    Following the advent of a network of upper-air Turner 1965) and that became a central theoretical is-
soundings over the western United States, Reed (1933) sue in much subsequent research.
recognized that the distribution of summer rains in this
region was, in fact, part of a larger-scale phenomenon.
The position of the “high-level anticyclone” east or 3. Moisture source debate
west of the Continental Divide, he proposed, dictated
the occurrence of rainfall events by stabilizing (west-                   Reitan (1957) and, later, Rasmusson (1967) raised
ward position) or destabilizing (easterly position) the several important points that seemed to contradict
lapse rate. Moreover, he believed advection of low- Bryson and Lowry’s (1955a,b) conclusions. Although
level moisture along the anticyclonic trajectory was not directly investigating moisture sources, Reitan
not necessary for destabilization because sufficient (1957)1 concluded that the greatest amount of precipi-
water vapor was presumably present throughout the table water vapor during the Arizona monsoon is found
region. Widespread convective activity was depen- below 800 mb. Considering the fact that the topogra-
dent, supposedly, on upper-level instability alone.                   phy between Arizona and the Gulf of Mexico rises
    The first comprehensive climatology of the NA above this level, the Gulf of Mexico appeared to be
monsoon phenomenon was presented by Bryson and an unlikely source for such low-level moisture. Simi-
Lowry (1955a, b). Although these authors were not the larly, Rasmusson (1967), in examining annual vapor
first to recognize the monsoonal nature of the circula-
tion regime (e.g., Ives 1949), the prominence of their 1Several citations of this source have been misleading. Reitan
work ensured that the NA monsoon would be recog- (1957) did not suggest the Gulf of California as an important
nized primarily as an Arizona summer feature, hence source of low-level moisture during the summer in Arizona. He,
the later predominance of Arizona in the literature. in fact, reiterates the findings of Bryson and Lowry (1955a) that
                                                                      the moisture comes from the Gulf of Mexico. After examining his
Bryson and Lowry (1955a, b) provided the dynamic conclusions, however, other authors raised the question of how
explanation for the sudden appearance of the summer Gulf of Mexico moisture could be advected over the Continental
thunderstorms that was lacking in Reed (1933). The Divide and still provide low-level moisture in Arizona.

2200                                                                                             Vol. 78, No. 10, October 1997
flux in North America for a 2-yr period, discovered a      west, gulf surges and more generally the lower atmo-
strong northward vapor flux into the southwestern          sphere over the Gulf of California have been a focus
United States from the Gulf of California and a much       of recent research (Douglas 1995; Douglas and Li
weaker inflow from east of the Rocky Mountains.            1996; Stensrud et al. 1997). The actual mechanisms
Taking these observations into account, the Gulf of        for surge initiation and their relationship to large-scale
California and eastern tropical Pacific arose again as     circulation patterns have been unclear (Hales 1972;
possible contributors to increases in moisture and con-    Brenner 1974). Recent evidence from Stensrud et al.
comitant rains during the NA monsoon.                      (1997) again points to the role of westward-moving
                                                           tropical disturbances south of the mouth of the Gulf
a. The gulf surge phenomenon                               of California (approximately 20°N) in the develop-
    Hales (1972) and Brenner (1974) argued that the        ment of major gulf surges. However, their findings
eastern tropical Pacific off the Mexican coast, and not    from mesoscale model simulations for the gulf region
the Gulf of Mexico, provides the greatest amount of        indicate that large-scale features can indeed be linked
water vapor for the NA monsoon west of the Conti-          to major surges. They contend that major surges oc-
nental Divide. To substantiate this hypothesis, they       cur when a midlatitude trough passes through west-
offered a mechanism by which atmospheric moisture          ern North America (around 40°N) preceding, by 1 or
is transported northward through the Gulf of Califor-      2 days, the passage of an easterly wave across the same
nia, occasionally as far inland as the Great Basin         longitudinal belt (Fig. 4). The increased subsidence
Desert. The transport mechanism was described as a
“surge” of moist tropical Pacific air that is channeled
up the Gulf of California as a result of a low-level at-
mospheric pressure gradient. This pressure gradient is
created as the thermal equilibrium between the tropi-
cal Pacific and the Gulf of California becomes dis-
rupted, primarily due to the development of a cloudy,
showery air mass at the mouth of the Gulf. The ther-
mal and, hence, pressure contrasts between the cooler,
showery air mass to the south and the intensely heated
surface in the deserts of Arizona and eastern Califor-
nia drive the moisture surge. A gulf surge is often
manifest as a strong shift to south–southeast winds
(particularly at Yuma, Arizona) accompanied by lower
temperatures and higher dewpoints, which are detect-
able at both upper-air stations (Empalme and
Mazatlán) adjacent to the Gulf of California. For the
most part, the gulf surge is a lower tropospheric fea-
ture, with all but the strongest surges typically con-
fined below the 700-mb level. The northward influx
of low-level maritime tropical air into the Sonoran
Desert and the uplands of Sonora and Arizona is be-
lieved to augment convective activity. Many other in-
vestigations have subsequently supported this mois-
ture surge hypothesis (Hales 1974; Houghton 1979;
Jáuregui and Cruz 1981; Tang and Reiter 1984;
Carleton 1985, 1986; Hasimoto and Reyes 1988;
Reyes and Cadet 1988; Carleton et al. 1990; Badan-
Dangon et al. 1991; Rowson and Colucci 1992; Dou-             FIG. 4. Conceptualization of the gulf surge phenomenon. Major
                                                           surges are associated with easterly waves crossing south of the
glas et al. 1993; Douglas 1995; Maddox et al. 1995;
                                                           Gulf of California, out of phase with an upper-level midlatitude
McCollum et al. 1995; Stensrud et al. 1997).               trough. Minor surges are associated with mesoscale convective
    Due to their apparent role in the convective envi-     systems drifting over the northern half of the Gulf (see text for
ronment of northwestern Mexico and the desert South-       details).

Bulletin of the American Meteorological Society                                                                      2201
The gulf surge does not, in itself, necessarily ac-
                                                                    count for the total Gulf of California–eastern Pacific
                                                                    moisture transport into the desert Southwest. Field
                                                                    studies (Badan-Dangon et al. 1991; Douglas 1995) de-
                                                                    scribe mean southerly flow at the northern end of the
                                                                    Gulf of California during July and August, with a low-
                                                                    level jet as a notable characteristic (see Fig. 5 and re-
                                                                    lated discussion in section 4a, also). Mesoscale obser-
                                                                    vations from SWAMP (35 days of pilot balloon sound-
                                                                    ings and several research aircraft measurements) docu-
                                                                    ment the nature of the low-level jet. It is present un-
                                                                    der various synoptic configurations, exhibiting an
                                                                    early morning maximum over the northern Gulf of
                                                                    California, with the jet appearing to be strongest un-
                                                                    der surge conditions (Douglas 1995). The persistence
                                                                    of this low-level flow suggests significant northerly
                                                                    transport of water vapor from the Gulf of California.
                                                                    For the Sonoran Desert region, a recent analysis
    FIG. 5. Schematic of selected dynamic features during the NA    (Schmitz and Mullen 1996) of atmospheric water va-
monsoon. The typical 850-mb late-afternoon height configuration     por flux (calculated for the time-mean wind during
(solid dark lines) includes heat lows over the elevated interior,   July and August) for the Gulf of California shows that
which act to draw in moisture by way of low-level jets (LLJ)
                                                                    time-mean moisture transport is dominant over that of
originating over the Gulf of California and the Gulf of Mexico
(large arrow). The early morning Gulf of California LLJ core is     transient flux (deviations from the mean), although the
shown via shaded isotachs (3, 5, and 7 m s−1) and streamlines at    role of transient flux may be more critical for actual
450 m (1500 ft) above ground level (adapted in part from Douglas    precipitation episodes over the region.
1995).
                                                                    b. A dual moisture source
                                                                       Despite the fact that the more recent literature has
behind the trough over the northern end of the gulf                 tended to view the tropical eastern Pacific–Gulf of
apparently creates conditions in surface pressure and               California as the dominant moisture source for the NA
low-level temperature fields conducive to surges. The               monsoon, the Gulf of Mexico is generally still viewed
actual initiation of major surges may result from cold              as an important contributor. Most investigations of the
outflows caused by deep convection over the Gulf of                 NA monsoon subscribe to the idea of moisture advec-
California, or it may simply result from the low-level              tion in some combination from both the eastern tropi-
convergence associated with the easterly wave.                      cal Pacific–Gulf of California and the Gulf of Mexico
   Minor gulf surges, those typically confined to the               (Hales 1974; Brenner 1974; Mitchell 1976; Carleton
northern half of the Gulf (Hales 1972), can be trig-                1985, 1986; Carleton et al. 1990; Harrington et al.
gered by the passage of a mesoscale convective com-                 1992; Watson et al. 1994a; Stensrud et al. 1995;
plex passing from the mainland over the Gulf, by                    Schmitz and Mullen 1996). Much recent work states
backdoor cold fronts approaching from the north to                  that low-level moisture is largely attributable to the
northeast as opposed to the typical west to northwest               Gulf of California and that upper-level water vapor is
direction (R. Maddox 1996, personal communication;                  transported from the Gulf of Mexico. The debate is
Hales 1972) or by the passage of an easterly wave                   more than academic because source regions have im-
without the preceding upper-level trough associated                 plications for seasonal forecasts (Schmitz and Mullen
with major surges (Stensrud et al. 1997), as concep-                1996) and for the development of the convective en-
tualized in Fig. 4. The weather effects for Sonora and              vironment (Stensrud et al. 1995).
Arizona of these minor surges are similar to major                     Carleton (1986) identifies contributions of both
surges; however, the former go undetected in the                    sources by examining a set of thermodynamic param-
Empalme and Mazatlán soundings, making forecast-                    eters for several Mexican and United States upper-air
ing these surges more difficult (R. Maddox 1996, per-               stations. Although the importance of the sources rela-
sonal communication).                                               tive to one another is not evaluated, he argues that rapid

2202                                                                                         Vol. 78, No. 10, October 1997
lower-level moisture surges moving up the Gulf of           Gall 1989; Carleton et al. 1990; Rowson and Colucci
California and slower high-level moisture from the          1992; Dunn and Horel 1994a,b; McCollum et al. 1995;
Gulf of Mexico work in concert during monsoon               Maddox et al. 1995). This may be a result of the fact
“bursts” (his term “burst” has the same connotation as      that the monsoon phenomenon was first recognized in
that for the Southwest Asian monsoon—i.e., large            the literature as the “Arizona summer monsoon”
surges of moisture and concomitant precipitation).          (Bryson and Lowry 1955a).2 Furthermore, it has been
Carleton et al. (1990) reaffirm this conclusion in their    stated that, in the United States, the NA monsoon is
examination of interannual variability of Arizona rain-     most visible in Arizona (Trewartha 1981; Adang and
fall during the NA monsoon. This recognition of two         Gall 1989; Carleton et al. 1990). Nevertheless, several
distinct moisture sources for the NA monsoon raises         investigators have expanded or shifted the geographic
important questions about the relative contributions of     focus of NA monsoon influence. Reyes and Cadet
the Gulf of Mexico and eastern tropical Pacific–Gulf        (1988) and Douglas et al. (1993) have shown that the
of California during monsoon bursts, as well as the         “Arizona monsoon” is merely an extension of a more
possible spatial and temporal variations in the relative    pronounced Mexican phenomenon centered over the
contribution from each source at different heights in       western foothills of the Sierra Madre Occidental (Fig.
the atmosphere. Maddox et al. (1995) claim that this        6). New Mexico has also been recognized as greatly
reduced role of the Gulf of Mexico as moisture source       influenced by the NA monsoon (Tuan et al. 1973;
is still too great, but the results of Schmitz and Mullen   Mitchell 1976; Trewartha 1981; Harrington et al.
(1996), focusing on the Sonoran Desert, suggest that        1992; Douglas et al. 1993; Bowen 1996), and one in-
the time-mean upper-level flow (700 mb to 200 mb)           dex of monsoon influence has shown that New Mexico
has its origin over the Gulf of Mexico, while low-level     is the state most affected by the NA monsoon (Dou-
flow is attributable to the Gulf of California. Clearly,    glas et al. 1993). For example, note the similarities in
this question has not been completely resolved.             precipitation regime between Tucson and Alamogordo
    One further complicating factor in the moisture         in Fig. 2. The NA monsoon moisture and attendant
source debate is the role of the Sierra Madre Occiden-      thunderstorms even extend into the lower Colorado
tal in moistening the midlevels of the atmosphere.          River valley and the Mojave Desert, and farther west
Recent empirical and modeling studies (Maddox et al.        to the transverse and peninsular ranges of southern
1995; Stensrud et al. 1995; Schmitz and Mullen 1996)        California, with some regularity (Tubbs 1972).
highlight the importance of low-level Gulf of Califor-          On a much broader scale, Tang and Reiter (1984)
nia and eastern Pacific moisture being drawn inland         and Reiter and Tang (1984) demonstrate that the west-
over the Sonoran and Sinaloan lowlands and up the           ern United States and northern Mexican plateau act to
western flanks of the Sierra Madre Occidental due to        create a true monsoonal circulation pattern over the
topographic heating. The resulting convection redis-        North American continent, comparable to that of the
tributes moisture to higher levels. This moistened air      Tibetan Plateau. This expanded region of summer
may then mix with elevated Gulf of Mexico moisture          monsoon influence, which they designate as the “west-
from the east and can subsequently be transported into      ern plateau,” ranges longitudinally from the Sierra
the southwestern United States via flow around the          Nevada in the west to the Colorado and Wyoming
anticyclone. As a result, the simple Gulf of Califor-       Rockies in the east. The northernmost limit runs from
nia–Gulf of Mexico moisture source dichotomy be-            Oregon along the Idaho–Utah border into Wyoming.
comes much more difficult to unravel.                       This limit, according to Tang and Reiter (1984), co-
                                                            incides with an anticyclonic shear line that marks the
                                                            northernmost extension of summer convective activ-
4. Geographic extent                                        ity. The Mexican Plateau along the international bor-
                                                            der serves as their southern boundary. Tang and Reiter
a. Spatial domain                                           (1984) postulate a low-level jet containing tropical
   It is quite apparent that Arizona has been the prin-     moisture drawn up the Gulf of California into the Great
cipal geographic study area for NA monsoon research,
particularly in relation to summer precipitation            2
                                                             Though this phenomenon was dubbed the Arizona monsoon,
(Bryson and Lowry 1955a,b; Reitan 1957; Bryson              Bryson (1957b) clearly recognized that this was also a Mexican
1957a,b; Green 1963; Carleton 1985, 1986, 1987;             feature and suggested that it should be properly called the
Moore et al. 1989; Smith and Gall 1989; Adang and           “Sonoran summer monsoon.”

Bulletin of the American Meteorological Society                                                                    2203
Basin desert of Nevada and Utah as a result of mon-
soonal pressure gradients. This corresponds closely to
ideas put forth by Brenner (1974), Hales (1974), and
Houghton (1979). Likewise, moisture is transported
east of the Continental Divide by a low-level jet from
the Gulf of Mexico. The moisture is moved up the
Continental Divide as a result of interior plateau heat-
lows, thereby increasing convective activity (Fig. 5).
These ideas are in agreement with the postulated dual
moisture source for the NA monsoon, and they indi-
cate that the Continental Divide may be important in
determining low-level moisture trajectories.

b. Distribution of air masses
   Several of the aforementioned studies have also
described the NA monsoon as a contrast between dry
and moist air masses in the southwestern United States
and northwestern Mexico. The moist air mass, char-
acterized as a moist tongue moving north from
Mexico into the southwestern United States, is
bounded to the north and west by drier air from the
Pacific subtropical high (Bryson and Lowry 1955a,b).
The zone of interaction between the air masses has
been viewed as a definite, but shifting, boundary that
exists between the moist air mass brought in by the
westward extension of the Bermuda high and the drier
eastern-Pacific air (Bryson and Lowry 1955a,b;
Mitchell 1976; Moore et al. 1989; Adang and Gall
1989; Watson et al. 1994a). For example, while the
transition from dry to moist air masses at the onset of
the monsoon season is not always rapid, it has been
shown to occur in as short a period as a few hours
(Nastrom and Eaton 1993). In addition, gradients in
temperature, water vapor content, and winds similar
to those of a quasi-stationary, midlatitude front be-
tween the moist tongue and dry air have also been
observed (Moore et al. 1989; Adang and Gall 1989).
The boundary of this moist tongue has been described
as shifting spatially in agreement with the westward-
expanding Bermuda high (Bryson 1957a; Trewartha
1981; Tang and Reiter 1984; Harrington et al. 1992),
from eastern New Mexico in June to western New
Mexico and central and eastern Arizona by July. East-
ern California and western Arizona are under its in-
fluence by August. Climatologically, this boundary

    FIG. 6. Frequency (in percent of total hours) of cloud-top
temperatures less than −38°C for June, July, and August of 1985–
92, derived from averaged hourly infrared GOES imagery
(adapted from Douglas et al. 1993). Shaded regions delineate areas
of systematic deep convection.

2204                                                                 Vol. 78, No. 10, October 1997
has been defined in terms of contrasting equivalent          rapidly to the west, across the Gulf of California along
potential temperature (Mitchell 1976). These studies         the Baja Peninsula and northward into California. The
of the NA monsoon boundary do not, however, ex-              greater influence of subsidence under the Pacific sub-
amine the specific contributions provided by the Gulf        tropical high in these latter areas has been suggested
of Mexico or the Gulf of California for the develop-         as a reason for the steep decrease in rainfall (Hastings
ment of the moist tongue.                                    and Turner 1965). Yet the drop-off takes place over a
                                                             relatively short distance (as compared to the scale of
                                                             subsidence); a concise, supported explanation for this
5. Spatial and temporal variability                          westward decrease is not found in the literature. North
                                                             of about 26°N on the Pacific side of the peninsula, the
   The NA monsoon region, like other arid and semi-          cool waters of the California current and strong sub-
arid locales, experiences great climatic variability on      sidence stifle summertime convection, and the winter
a range of spatial and temporal scales. Difficulties in      precipitation regime of the West Coast dominates
understanding the variability of summertime convec-          (Hastings and Turner 1965). These patterns of precipi-
tive activity in the southwestern United States and          tation seasonality are illustrated in Fig. 2.
northwestern Mexico result from the complex inter-              Within the United States, southern and eastern Ari-
actions between synoptic and mesoscale circulation           zona, much of New Mexico, and south-central Colo-
features and the extremely varied topography. Al-            rado receive precipitation related to the NA monsoon,
though larger-scale atmospheric motions control the          but rainfall amounts are much more variable than in
broad distribution of water vapor and the general sta-       northwestern Mexico. For example, the Mogollon Rim
bility or instability in the atmosphere, local topo-         of central Arizona and the White Mountains of east-
graphic effects in the western United States and north-      ern Arizona receive a much greater percentage of rainy
western Mexico are critical to the geographic and even       days during the monsoon than the lower desert to the
temporal distribution of convective activity. In fact,       southwest, despite generally higher dewpoints in this
dealing with these factors has been at the heart of many     latter region (although relative humidities are lower).
modeling and forecasting efforts in the region, as we        This is not unexpected considering that convection is
discuss in section 6b.                                       also greater over the elevated regions of Arizona.
                                                             Across the NA monsoon area, regional-scale spatial
a. Regional-scale spatial variability                        variability in precipitation is closely linked to the pen-
   Spatial variability of the NA monsoon has been            etration of moisture into the interior regions (e.g.,
examined almost exclusively in terms of precipitation.       Hales 1972; Brenner 1974), but synoptic-scale con-
The geographic influence of NA monsoon rains is pri-         trols are believed to be equally important for the re-
marily centered in the western foothills of the Sierra       gional variability of NA monsoon precipitation in
Madre Occidental in northwestern Mexico, specifi-            Arizona (Carleton 1986). Although not directly part
cally in the Mexican states of Nayarit, Sinaloa, and         of the NA monsoon, decaying tropical cyclones also
Sonora (Hales 1972; Douglas et al. 1993). This region        contribute to the regional variation of total summer
receives up to 70% of its annual rainfall in the months      precipitation (Reyes 1988; Hereford and Webb 1992),
of July, August, and September (Mosiño and Garcia            particularly over Sinaloa, Sonora, and the southern tip
1974; Douglas et al. 1993; Reyes et al. 1994). The in-       of Baja California.
tensity of convection over the Sierra Madre is appar-
ent from satellite imagery (Douglas et al. 1993; Negri       b. Local-scale variability
et al. 1993; Stensrud et al. 1995), as illustrated in Fig.      Much of the early NA monsoon research focused
6. The distinctness of the July through September            on the spatial and temporal variability of Arizona sum-
maximum decreases rapidly as one moves northward             mer rainfall (McDonald 1956; Bryson 1957a; Green
through Sonora toward the international border. The          1960). Considering the nature of air mass thunder-
NA monsoon precipitation regime also becomes less            storms, it is not surprising that there is a great spatial
notable toward the eastern flank of the Sierra Madre         variability in precipitation and a variable geographic
Occidental and onto the Mexican Altiplano. Much of           distribution of summer thunderstorms. In Arizona,
northeastern Mexico actually experiences a June and          only small-to-moderate interstation correlations exist
September rainfall maximum with a minimum in July            between rainfall events (McDonald 1956). Yet, diur-
and August. Similarly, NA monsoon rains diminish             nal variability in summer thunderstorms does show a

Bulletin of the American Meteorological Society                                                                 2205
evated cool layer is supposedly advected west and
                                                                  southwest over the low desert toward evening, thereby
                                                                  destabilizing the local environment (although no fur-
                                                                  ther support of this idea has been found). Ascending
                                                                  air in the late evening and early morning over the
                                                                  Sonoran Desert low may also support the nocturnal
                                                                  convection (Douglas and Li 1996), as a result of the
                                                                  larger lapse rate. With regard to the development of
                                                                  severe nocturnal mesoscale convective systems (MCS)
                                                                  over southern and central Arizona, McCollum et al.
                                                                  (1995) state that subsynoptic scale features are also of
                                                                  great import, particularly low-level moisture advection
                                                                  from the Gulf of California and localized convergence
                                                                  in favored areas.

                                                                  c. Seasonal variability
                                                                     Apart from the spatial variability that is directly
                                                                  related to topographic features, there is much intra-
                                                                  seasonal variability of NA monsoon rainfall (in terms
                                                                  of intensity and areal extent), particularly in the south-
                                                                  western United States. The highlands of eastern
                                                                  Sinaloa and Sonora in northwestern Mexico experi-
   FIG. 7. Percentage of precipitation events occurring between   ence shower activity almost every day during the en-
2000 and 0800 true solar time in Arizona for the months of July   tire NA monsoon season. Nevertheless, there are ap-
and August [redrafted from Balling and Brazel (1987)]. Note
gradient from the Mogollon Rim and Colorado Plateau to the low
                                                                  parent periodicities in rainfall episodes in different
deserts of central Arizona.                                       areas of northwestern Mexico, which may be attribut-
                                                                  able to variations in source region influence (Reyes et
                                                                  al. 1994). Intraseasonal variation of precipitation to-
marked geographic dependence, particularly in Ari-                tals in this region may be related to the frequency of
zona (McDonald 1956; Ackerman 1959; Sellers 1958;                 MCS and tropical perturbations that account for a great
Hales 1972, 1977; Brenner 1974; Reiter and Tang                   portion of the yearly precipitation (Reyes et al. 1994).
1984; Balling and Brazel 1987; Maddox et al. 1991;                Farther north, much of the intraseasonal precipitation
Watson et al. 1994b; Maddox et al. 1995). Diurnal                 variability may be related to the gulf surge phenom-
convective activity is generally at a maximum over the            enon described by Hales (1972) and Brenner (1974).
Colorado Plateau and the higher elevations of the                 These surges are believed to contribute significantly
Continental Divide during early afternoon. Over the               to the widespread summer rainfall events over much
highlands of southern Arizona and northern Sonora,                of the desert Southwest. Considering that several gulf
maximum activity is characteristic of the early evening           surges can occur in this area during a given NA mon-
hours. By late evening and early morning, the coastal             soon season, they may account for a great deal of the
lowlands of northwest Mexico and the lower desert                 variability in total summer precipitation (Brenner
regions of southern and central Arizona experience a              1974). It is important to note here that gulf surges,
nighttime convective maximum (in contrast to the tim-             weak or strong, have been assumed to result in in-
ing of convective storms elsewhere), as illustrated in            creased convective activity and precipitation, but to
Fig. 7. Similar diurnal precipitation variation due to            date no study has been published in the formal litera-
topographic effects has also been noted in central New            ture quantifying this relationship.
Mexico (Bowen 1996). The specific causes of the noc-                 A number of authors have examined changes in the
turnal maximum have been explored but are still not               synoptic-scale circulation and its effects on intra-
clearly understood. Hales (1977) states that midtropo-            seasonal variability of the NA monsoon (Bryson and
spheric cooling results from localized evaporation of             Lowry 1955a, b; Carleton 1986, 1987). Analyses of
cumulonimbus towers that develop over the higher                  composite synoptic pressure patterns have identified
terrain of central Arizona in the afternoon. This el-             atmospheric configurations that enhance or stifle

2206                                                                                       Vol. 78, No. 10, October 1997
convective activity (bursts or
breaks) in terms of cloudiness,
precipitation, and cloud-
to-ground lightning (Carleton
1986, 1987; Watson et al.
1994b). Carleton (1986, 1987)
has shown that widespread con-
vective activity in the south-
western United States is typi-
cally associated with passing
upper-level troughs in the west-
erlies. In addition, the northward
displacement of the subtropical
ridge and the formation of a cut-
off “four corners high” also re-
sults in increased thunderstorm
activity in Arizona. A southerly
displacement of the subtropical
ridge over northern Mexico
tends to result in (drier) break
conditions over the desert South-
west. His conclusions highlight         FIG. 8. Composites of 500-mb geopotential height (dam), 1946–88, showing the overall
the importance of the location of mean circulation for July and August (upper left), the means of the five wettest and driest
the subtropical ridge in control- years in Arizona (upper right; solid contour lines correspond to wet years, and broken lines
ling convective activity in the correspond to dry years), and the wet (lower left) and dry (lower right) anomalies from the
                                     overall mean (precipitation data are from the seven Arizona climate divisions).
southwestern United States.
In examining cloud-to-ground
lightning, Watson et al. (1994a) found that similar syn- d. Interannual variability
optic configurations to those of Carleton led to bursts           The sparseness of long-term instrumental precipi-
and breaks in total convective activity. Again, a south- tation records (both temporally and spatially) in the
erly shift of the subtropical ridge axis places Arizona southwestern United States and northwestern Mexico
under the strong, drying effects of the westerlies, has hindered research on NA monsoon variability at
which promote break conditions. A northerly shift of interannual timescales. Nonetheless, several studies
the subtropical high is associated with increased in- have noted marked variations in variability of the NA
flow of moisture from the south and results in wide- monsoon since the turn of the century (Green and Sell-
spread shower activity.                                        ers 1964; Bradley 1976; Tenharkel 1980; Brazel and
   Although the results of the studies by Carleton Prasad 1984; Carleton et al. 1990; Hereford and Webb
(1986, 1987) and Watson et al. (1994a) agree on the 1992).
importance of the latitudinal position of the subtropi-           Interannual variability in NA monsoon activity may
cal ridge in determining bursts and breaks, Maddox result from the corresponding variability of certain
et al. (1995) point out that the composite 500-mb burst synoptic-scale patterns, similar to those associated
pattern of Watson et al. (1994a; their Fig. 18c) closely with intraseasonal variability. Carleton et al. (1990)
resembles Carleton’s (1986) composite break pattern have argued that latitudinal shifts in the midlevel sub-
(his Fig. 5b). Their definition of burst and breaks are tropical ridge over the southwest United States, which
based, however, on different criteria: Carleton (1986) accounts for a great deal of within-season monsoon
on total cloudiness over a large portion of the western variability, are also responsible for year-to-year and
United States, and Watson et al. (1994a) on cloud-to- decadal-scale variability. Figure 8 shows that, in gen-
ground lightning only in Arizona. Such intricacies eral, northerly displacement of the subtropical ridge
further demonstrate the complex nature of the relation- is associated with wetter Arizona summers, while a
ship between convective activity and synoptic circu- southerly shift coincides with decreases in summer
lation during the NA monsoon.                                  rainfall totals (there are exceptions; e.g., 1990 was a

Bulletin of the American Meteorological Society                                                                        2207
very wet year even though there was a southerly shift).     eastern North America delayed the development of the
Carleton et al. (1990) found that latitudinal displace-     Bermuda high and associated summer rains west of the
ment of the subtropical ridge is tied to different phases   Continental Divide (van Devender and Spaulding
of the Pacific–North American (PNA) teleconnection          1979). The period since the establishment of a mon-
pattern, which in turn is linked to the El Niño–South-      soonlike circulation has included intervals consider-
ern Oscillation (ENSO). Wetter summers are associ-          ably wetter and drier than those found in the instru-
ated with the meridional phase of the PNA and posi-         mental record (Davis 1994; Petersen 1994). In general,
tive Pacific SST anomalies that enhance ridging over        the atmospheric controls on the paleomonsoon are
North America. Drier summers have tended to follow          assumed to be fundamentally the same as those for the
zonal phases of the PNA.                                    modern monsoon (i.e., position and strength of sub-
                                                            tropical highs and associated latitudinal position of the
e. ENSO-related studies                                     westerlies), with changes in their relative location and
   Relationships between ENSO and NA monsoon                intensity causing the observed long-term variability.
variability have been examined for possible causality,
but such links have proven to be elusive. Some stud-
ies have analyzed the strength and geographic distri-       6. Severe storms
bution of rainfall during and following ENSO events.
More frequent tropical perturbations off the southwest-     a. Convective environment
ern coast of Mexico have tended to follow the mature           An important aspect of NA monsoon research deals
ENSO stage (Northern Hemisphere winter) when the            with severe storms and their predictability. The south-
ITCZ is displaced farther north and there is increased      western United States and northwestern Mexico have
northward transport of water vapor (Reyes and Cadet         experienced tremendous population growth in the last
1988; Reyes and Mejía-Trejo 1991). Under these con-         few decades, and severe thunderstorms and related
ditions, parts of northwestern Mexico (particularly the     phenomena can pose a serious hazard for residents of
Sonoran Desert) have experienced positive anomalies         this region. A range of studies have been performed,
in summer rainfall (Reyes and Mejía-Trejo 1991).            including those on the severe storm environment
Within the southwestern United States, Andrade and          (Hales 1975; Randerson 1986; Smith and Gall 1989;
Sellers (1988) found little correlation between ENSO        McCollum et al. 1995; Maddox et al. 1995), lightning
and total summer rainfall in New Mexico and Arizona.        (Reap 1986; Watson et al. 1994a; Watson et al. 1994b),
Yet Harrington et al. (1992) found a correlation be-        dust storms (Idso et al. 1972; Nickling and Brazel
tween ENSO warm and cold events and the geographic          1984; Brazel and Nickling 1986), and flash flooding
distribution of summer precipitation for the same ar-       (Randerson 1976, 1986; Maddox et al. 1980). In re-
eas. Likewise, Hereford and Webb (1992) suggest             cent years, there has been a greater focus on the con-
there is some relationship between increased summer         vective environment of severe storms in the NA mon-
precipitation in the Colorado plateau region and the        soon. Much of the latest research has resulted from the
warm ENSO phase. Thus, relationships between                SWAMP, which is aimed at greater understanding of
ENSO-related phenomena and the NA monsoon are               convective environments in central Arizona and north-
not yet well understood.                                    western Mexico (Maddox 1990; Meitin 1991; Reyes
                                                            et al. 1994).
f. Paleoclimate studies                                        Two common approaches for severe storm analy-
   The focus of this review is on the modern NA mon-        sis in the NA monsoon region have been analysis of
soon, but a number of studies utilizing proxy data in-      synoptic patterns (Maddox et al. 1980; Randerson
clude description of summer paleoclimate in the south-      1976; Brazel and Nickling 1986; Maddox et al. 1995)
western United States (e.g., Wells 1979; van Devender       and of satellite imagery (Hales 1975; Smith and Gall
and Spaulding 1979; Davis and Shafer 1992). Briefly,        1989; Farfán and Zehnder 1994). Smith and Gall
paleoenvironmental conditions consistent with a cir-        (1989) describe a characteristic MCS of the NA mon-
culation similar to the modern NA monsoon were              soon that shares a number of characteristics with tropi-
present about 10 000 to 8000 BP, or roughly 10 000          cal squall lines. Here, MCSs typically form over the
yr after the last glacial maximum (van Devender and         elevated regions of southeastern Arizona and north-
Spaulding 1979; Davis and Shafer 1992). Prior to this       ern Sonora as the result of the merging of late after-
time, the presence of the Laurentide ice sheet over         noon thunderstorm cells. By evening, the storms tend

2208                                                                                 Vol. 78, No. 10, October 1997
FIG. 9. Example of a severe thunderstorm over central Arizona showing the strong ridge over the Utah–Nevada border and associated
NE flow over Arizona, represented by (a) the 500-mb geopotential height pattern (dam) for 0000 UTC 10 August 1984 (1700 LST 9
August 1984) with matching wind barbs (m s−1) from radiosonde observations, and (b) GOES-W infrared image for 0100 UTC 10
August 1984 (1800 LST 9 August 1984). The broken line in (a) frames the image in (b).

to move in a westerly direction toward the low desert              been identified as precursors to severe thunderstorm
via discrete propagation. Although their life span is              development in central Arizona (Dunn and Horel
typically less than 1 day, they often produce remnant              1994a; Maddox et al. 1995), and although these pat-
cyclonic circulation that can stimulate convection the             terns are not always positioned directly over Arizona,
next day over southern California and Baja Califor-                they do provide clues for the evolution of the central
nia. Interestingly, this behavior is not too far from the          Arizona convective environment (Maddox et al. 1995).
turn-of-the-century description of the “Sonora storm”              Yet subsynoptic features such as low-level moisture
(Campbell 1906). The necessary pattern for the for-                advection from the Gulf of California and localized
mation of these squall lines appears to be a 500-mb                convergence resulting from the complex terrain of
ridge with its axis centered over the Utah–Nevada                  central Arizona are also critical for development of the
border. This configuration promotes strong flow from               severe thunderstorm environment. As a result, fore-
the east and northeast over the elevated terrain along             casting of severe thunderstorms for this region is ex-
the Continental Divide. Under these conditions, dry                tremely difficult (Dunn and Horel 1994a,b; McCollum
midtropospheric air is advected westward over moist                et al. 1995).
warm air covering the lower terrain of Sonora and
southern Arizona. If low-level wind shear is present,              b. Forecasting and modeling
squall lines may form. Work by Farfán and Zehnder                     Because the convective environment is tied to par-
(1994) highlights the association between strong                   ticular synoptic patterns, to various subsynoptic fea-
midlevel easterlies over the Sierra Madre Occidental               tures, and to complex topography, considerable diffi-
in the formation of propagating MCS over northern                  culties must be overcome for successful model predic-
Mexico.                                                            tion and forecasting. A major problem is that nearly
   The same anomalous ridge pattern has also been                  identical synoptic situations can produce days where
associated with severe storms in central Arizona                   thunderstorms are either confined to topographically
(Maddox et al. 1995) and dust storm generation in                  favored locations or where thunderstorm activity
Arizona (Brazel and Nickling 1986). Figures 9a and                 spreads to lower elevations (Dunn and Horel 1994a).
9b illustrate an example of this pattern for a severe                 In Arizona, the forecasting skill scores for predic-
storm over central Arizona. Other patterns have also               tion of severe storm events have been exceptionally

Bulletin of the American Meteorological Society                                                                             2209
low in comparison to other regions of the continental      the westward expansion of the Bermuda high was rec-
United States (Maddox et al. 1995). Likewise, numeri-      ognized early on, following the availability of upper-
cal weather prediction models have not performed           air data. Synoptic climatological studies have provided
well for summer convection in Arizona. Dunn and            further details on the subtropical ridge and the effects
Horel (1994a,b) describe the difficulties experienced      of its latitudinal displacement. More recently, topo-
with the National Center for Environmental Predic-         graphic influences and mesoscale phenomena have
tion models (both the nested grid and Eta models) in       been highlighted as critical components of the NA
predicting southwest United States precipitation.          monsoon circulation. These findings have had impor-
Overall, both models provided poor predictions of          tant consequences for the debate over moisture
heavy rain episodes over Phoenix. Several mesoscale        sources. Most investigators are now in agreement that
features prevented the Eta model from properly simu-       low-level moisture is advected from the eastern tropi-
lating severe convection over Phoenix. The model           cal Pacific–Gulf of California, while the Gulf of
does not resolve the advection of low-level moisture       Mexico may contribute to upper-level moisture. In
from the Gulf of California into central Arizona,          fact, recent work suggests that moisture from these two
which is critical for destabilization of the local envi-   sources becomes mixed over the Sierra Madre Occi-
ronment prior to convection. This may result from the      dental, prior to being transported northward into the
failure of the model to produce the low-level jet above    United States.
the Gulf of California that seems to be responsible for       The spatial extent of the NA monsoon is large, cov-
moisture transport into the low deserts (Douglas et al.    ering much of the western United States and north-
1993; McCollum et al. 1995). This, in turn, may re-        western Mexico. Historically, most of the literature has
sult from inaccurate initial conditions for summertime     focused on Arizona, but the NA monsoon is actually
SSTs over the Gulf of California (Stensrud et al. 1995).   most spatially consistent over northwestern Mexico
In addition, another critical factor in localized severe   with greater variability to the north. Variability at a
convection is that of thunderstorm outflows (McCollum      range of temporal scales is related to interactions be-
et al. 1995), which are not included in the model’s        tween synoptic and mesoscale circulations, in particu-
convection scheme. Stensrud et al. (1995) found that       lar the latitudinal position of the subtropical ridge and
many of the general features of the NA monsoon cir-        the low-level moisture surges from the Gulf of Cali-
culation could be reproduced using the PSU–NCAR            fornia. Interannually, these relationships are partly in-
mesoscale model. They demonstrate that model ini-          fluenced by the PNA pattern (and perhaps to some
tialization of mesoscale features such as the low-level    degree by ENSO). Forecasting of the NA monsoon,
winds over the Gulf of California and the summertime       especially the ability to distinguish between weak or
warming of the Gulf are critical for dynamically con-      strong severe storm outbreaks, has been poor because
sistent simulations of mesoscale structures.               current models do not adequately resolve the mesos-
                                                           cale features. Recent research from the SWAMP
                                                           project, together with mesoscale modeling, has further
7. Summary and conclusions                                 highlighted the integrated roles of synoptic circulation,
                                                           topography, and mesoscale circulations in controlling
   The NA monsoon is one of the more complex and           NA monsoon activity.
interesting atmospheric circulation features over North       Many of the relationships noted above are not yet
America, and the associated research literature dates      well understood, and there are a number of important
back to the early part of this century. Although the       questions for future research.
circulation covers a significant portion of the conti-
nent, research on the NA monsoon (as well as its           1) What are the relative roles of convective mixing
broader understanding in the meteorological and cli-          over the Sierra Madre Occidental and subsequent
matological communities) has not enjoyed a match-             northward moisture transport compared to low-
ing prominence. The recent appearance of several pa-          level moisture from the Gulf of California and
pers from the SWAMP project has, however, some-               upper-level moisture from the Gulf of Mexico,
what improved this situation, especially in the context       with respect to widespread rainfall across the NA
of its mesoscale manifestations.                              monsoon region?
   It appears that a consensus has gradually emerged       2) Additionally, what are the spatial and temporal
regarding major circulation mechanisms. The role of           variations in each of these kinds of moisture de-

2210                                                                                Vol. 78, No. 10, October 1997
You can also read